| 1 |
:orphan: |
| 2 |
|
| 3 |
Current Clamp [IClamp] Stimulation |
| 4 |
================================== |
| 5 |
|
| 6 |
One of the most common way to provide stimulus into a network of cells is to directly inject a current |
| 7 |
into all or a subset of cells. This can be easily done in bmtk using the "**inputs**" section of the |
| 8 |
`SONATA simulation config <https://alleninstitute.github.io/bmtk/simulators.html#creating-the-environment>`_. |
| 9 |
For example, to provide an single 0.15 pA injection into all cells starting at 1000 ms and ending at 1500 ms; |
| 10 |
open the simulation *json* configuration file with your prefered text-editor and add the following subsection |
| 11 |
to the **inputs** section: |
| 12 |
|
| 13 |
.. code:: json |
| 14 |
|
| 15 |
{ |
| 16 |
"inputs" { |
| 17 |
"current_clamp": { |
| 18 |
"module": "IClamp", |
| 19 |
"input_type": "current_clamp", |
| 20 |
"node_set": "all", |
| 21 |
"amp": 0.1500, |
| 22 |
"delay": 1000.0, |
| 23 |
"duration": 500.0, |
| 24 |
"section_name": "soma" |
| 25 |
} |
| 26 |
} |
| 27 |
} |
| 28 |
|
| 29 |
Here "current_clamp" is the name we arbitarily give for this particular input, **module** and **input_type** are |
| 30 |
specify the type of stimulus we are using, and the other parameters are used to define the attributes of our |
| 31 |
current clamp (see parameter options below). |
| 32 |
|
| 33 |
After saving the file and run the simulation we should see the effect of a direct current injection into the |
| 34 |
soma of our cell model. |
| 35 |
|
| 36 |
.. figure:: _static/images/iclamp_stim_single.png |
| 37 |
:scale: 100% |
| 38 |
|
| 39 |
|
| 40 |
We can also have the option of add multiple independent current clamps in the same simulation, where each one |
| 41 |
may have a different injection trace, different onset and offset, target different subset of cells or be placed |
| 42 |
at different locations. For example to apply a series of three different current injections with an increasing |
| 43 |
amplitude we can just update the json file accordingly |
| 44 |
|
| 45 |
.. code:: json |
| 46 |
|
| 47 |
{ |
| 48 |
"inputs" { |
| 49 |
"current_clamp_1": { |
| 50 |
"module": "IClamp", |
| 51 |
"input_type": "current_clamp", |
| 52 |
"node_set": "all", |
| 53 |
"amp": 0.1500, |
| 54 |
"delay": 500.0, |
| 55 |
"duration": 500.0, |
| 56 |
"section_name": "soma" |
| 57 |
}, |
| 58 |
"current_clamp_2": { |
| 59 |
"module": "IClamp", |
| 60 |
"input_type": "current_clamp", |
| 61 |
"node_set": "all", |
| 62 |
"amp": 0.1750, |
| 63 |
"delay": 1500.0, |
| 64 |
"duration": 500.0, |
| 65 |
"section_name": "soma" |
| 66 |
}, |
| 67 |
"current_clamp_3": { |
| 68 |
"module": "IClamp", |
| 69 |
"input_type": "current_clamp", |
| 70 |
"node_set": "all", |
| 71 |
"amp": 0.2000, |
| 72 |
"delay": 2500.0, |
| 73 |
"duration": 500.0, |
| 74 |
"section_name": "soma" |
| 75 |
} |
| 76 |
} |
| 77 |
} |
| 78 |
|
| 79 |
.. figure:: _static/images/iclamp_stim_multiple.png |
| 80 |
:scale: 100% |
| 81 |
|
| 82 |
|
| 83 |
The above is a sequence of different current clamps done in series, but it is possible to have multiple current |
| 84 |
clamps run in parallel too. |
| 85 |
|
| 86 |
|
| 87 |
Advanced Stimulus Inputs |
| 88 |
------------------------ |
| 89 |
|
| 90 |
Where there are only a few step currents it is easy enough to use the above method of adding mutliple current |
| 91 |
clamps in series or in parallel. However for more advanced patterns of stimuli - like randomized noise, ramps, |
| 92 |
sine waves, etc. - the above method becomes impracticle. Instead of just passing in a single step-block, BMTK |
| 93 |
provides methods for modifying the **input_type** to pass in lists or even external files. |
| 94 |
|
| 95 |
List of Amplitudes |
| 96 |
++++++++++++++++++ |
| 97 |
|
| 98 |
In the above example, instead of using three separate current clamps, we can instead specify just a single one |
| 99 |
whose amplitudes change over time: |
| 100 |
|
| 101 |
.. code:: json |
| 102 |
|
| 103 |
{ |
| 104 |
"inputs" { |
| 105 |
"current_clamp_list": { |
| 106 |
"module": "IClamp", |
| 107 |
"input_type": "current_clamp", |
| 108 |
"node_set": "biophys_cells", |
| 109 |
"amp": [0.1500, 0.1750, 2500.0], |
| 110 |
"delay": [500.0, 0.1500.0, 2500.0], |
| 111 |
"duration": [500.0, 500.0, 500.0] |
| 112 |
} |
| 113 |
} |
| 114 |
} |
| 115 |
|
| 116 |
**Note** |
| 117 |
* Here the "**duration**" is optional and used to set the injection current back to 0. If not specified BMTK will |
| 118 |
just update the **amp** value of the current injection at the next **delay** |
| 119 |
* In many siutations like the one above, replace three separate current clamps with a list will produce the same results, however |
| 120 |
it is not guarenteed. |
| 121 |
|
| 122 |
Storing Amplitude Traces in a CSV file |
| 123 |
++++++++++++++++++++++++++++++++++++++ |
| 124 |
|
| 125 |
For even more complex current injection traces, users can create and pass in a csv file of the trace. The csv file |
| 126 |
should have two columns, one containing the changes in amplitudes and a corresponding one for the respective onsets, labeled |
| 127 |
**amps** and **timestamps** respectivly (although this headers can be changed, see options below) with a space delimiter |
| 128 |
(which can be changed using options below). |
| 129 |
|
| 130 |
|
| 131 |
.. list-table:: custom_current_traces.csv |
| 132 |
:widths: 25 25 |
| 133 |
:header-rows: 1 |
| 134 |
|
| 135 |
* - timestamps |
| 136 |
- amps |
| 137 |
* - 800.0 |
| 138 |
- 0.1500 |
| 139 |
* - 810.0 |
| 140 |
- 0.1600 |
| 141 |
* - 820.0 |
| 142 |
- 0.1700 |
| 143 |
* - 830.0 |
| 144 |
- 0.1800 |
| 145 |
* - ... |
| 146 |
- ... |
| 147 |
|
| 148 |
Then once the file has been saved update the json configuration. To do so we must make sure the **input_type** option is |
| 149 |
set to *csv*, and instead of options like `amp`, `delay`, and `duration` we just need to point bmtk to the `file` path of |
| 150 |
our csv: |
| 151 |
|
| 152 |
.. code:: json |
| 153 |
|
| 154 |
{ |
| 155 |
"inputs" { |
| 156 |
"current_clamp_csv": { |
| 157 |
"module": "IClamp", |
| 158 |
"input_type": "csv", |
| 159 |
"node_set": "all", |
| 160 |
"file": "custom_current_trace.csv" |
| 161 |
} |
| 162 |
} |
| 163 |
} |
| 164 |
|
| 165 |
Using Allen Cell-Types Database Sweeps |
| 166 |
++++++++++++++++++++++++++++++++++++++ |
| 167 |
|
| 168 |
In the `Allen Cell-Types Database <https://celltypes.brain-map.org/data>`_ there is the option to download the ephys |
| 169 |
nwb files which include "sweeps" of current injection used to classify various cells and optimize the model parameters. |
| 170 |
|
| 171 |
.. figure:: _static/images/ctdb_sweep_selection.png |
| 172 |
:scale: 100% |
| 173 |
|
| 174 |
Using the `Download data` button will download a nwb file containing various "sweeps" (as can be seen using the |
| 175 |
`select stimulus type` and `Select sweep` buttons). You can recreate these sweeps with BMTK using this downloaded file. |
| 176 |
As before, we need to update the *json* simulation configuration file, this time setting **input_type** to a value |
| 177 |
`allen` or `nwb`, specify the `file` path of the nwb, and the `sweep` number to use. |
| 178 |
|
| 179 |
.. code:: json |
| 180 |
|
| 181 |
{ |
| 182 |
"inputs" { |
| 183 |
"current_clamp_nwb": { |
| 184 |
"module": "IClamp", |
| 185 |
"input_type": "allen", |
| 186 |
"node_set": "all", |
| 187 |
"file": "/626159623_ephys.nwb", |
| 188 |
"sweep_id": 29, |
| 189 |
} |
| 190 |
} |
| 191 |
} |
| 192 |
|
| 193 |
IClamp parameters |
| 194 |
----------------- |
| 195 |
**module \: string** |
| 196 |
Should always be set to a value `IClamp` |
| 197 |
|
| 198 |
**input_type \: string** |
| 199 |
The format of the input, options: `current_clamp`, `csv`, `file` (same as `csv`), `allen`, or `nwb` (same as `allen`) |
| 200 |
|
| 201 |
**node_set \: string**, **dict**, or **list** |
| 202 |
Specify the subset of cells which the injection current will be applied. The default value of `all` will |
| 203 |
input the same current clamp to "all" cells in the network. Alternatively, you can filter by properties: |
| 204 |
|
| 205 |
.. code:: python |
| 206 |
|
| 207 |
"node_set": { |
| 208 |
"population": "v1" |
| 209 |
"location": "VisL4", |
| 210 |
"model": "PValb" |
| 211 |
} |
| 212 |
|
| 213 |
|
| 214 |
Or pass in a list of node-ids |
| 215 |
|
| 216 |
.. code:: python |
| 217 |
|
| 218 |
"node_set": [0, 100, 200, 300, 400] |
| 219 |
|
| 220 |
|
| 221 |
**section_name [optional] \: string** |
| 222 |
For morphologically detailed cells, the section of the cell that the current clamp will be placed. options |
| 223 |
include `soma`, `axon`, `dend`, `apic`, `other`; default value `soma` |
| 224 |
|
| 225 |
**section_index [optional] \: integer** |
| 226 |
For morphologically detailed cells, the NEURON designated branch of the given section where the current clamp will be placed. |
| 227 |
|
| 228 |
**section_dist [optional] \: float [0, 1)** |
| 229 |
For morphologically detailed cells, the distance along the branch where the current clamp will be placed. |
| 230 |
|
| 231 |
**file \: string** |
| 232 |
When using a `csv` or `nwb` **input_type**, path the file containing current clamp trace (or sweep) and is **[required]**. |
| 233 |
|
| 234 |
|
| 235 |
| |
| 236 |
|
| 237 |
*Parameters applicable only to current_clamp input_type* |
| 238 |
|
| 239 |
**amp \: float**, **list of floats** |
| 240 |
Amplitudes (pA) of current injection. |
| 241 |
|
| 242 |
**delay \: float**, **list of floats** |
| 243 |
Current injections onsets (ms) started from simulation time 0.0 |
| 244 |
|
| 245 |
**duration \: float**, **list of floats** |
| 246 |
Current injections durations (ms). |
| 247 |
|
| 248 |
|
| 249 |
| |
| 250 |
|
| 251 |
*Parameters applicable only to csv input_type* |
| 252 |
|
| 253 |
**separator [optional] \: string** |
| 254 |
csv file delimiter used to distinguish columns; defaul " " (single space) |
| 255 |
|
| 256 |
**timestamps_column [optional] \: string** |
| 257 |
Header name for column in csv used to specify changes in amplitude values; default `timestamps` |
| 258 |
|
| 259 |
**amplitudes_column [optional] \: string** |
| 260 |
Header name for column in csv used to specify ampltidue changes at each given timestep; default `amps` |
| 261 |
|
| 262 |
|
| 263 |
| |
| 264 |
|
| 265 |
*Parameters applicable only to allen input_type* |
| 266 |
|
| 267 |
**sweep_id \: integer** |
| 268 |
The sweep number of the wanted stimulus pattern found in the nwb file |
| 269 |
|
| 270 |
**downsample [optional] \: float** |
| 271 |
When specified, will attempt to downsample the injection current trace. As a good practice it is optimal |
| 272 |
to set the value to the same as the `dt` in the simulation configuration. Since for most Cell-Type sweeps |
| 273 |
the rate of current injection changes is typically much smaller than the simulation `dt`. |
| 274 |
|
| 275 |
**sweep_window [optional] \: [float, float]** |
| 276 |
Instead of using the full sweep, which is often longer than the simulation time, selection a portion of |
| 277 |
the sweep time from starting specifying a `0 <= start-time < stop-time < sweep-total-time` (in ms). |